A Study of Ex Ante Law Enforcement in Norm-Governed Learning Agents

摘要

We investigate ex ante law enforcement within a population of norm-governed learning agents using a probabilistic rule-based argumentation framework. We show that this formal framework can advantageously complete a traditional analysis based on expected utilities, in particular when hyper-rational or omniscient agents are not assumed. This has significant implications for the design of self-organising electronic institutions, where the cost of monitoring and enforcement of laws and norms has to be taken into consideration.